Library MCP Server

by siftdex Your server? Claim it

answering

Library is answering right now. Last checked 11 min ago. It exposes 13 tools.

Verified doc corpora for agents: grep-first retrieval, hashed pages, Merkle+RFC-3161 receipts

Uptime history 91 days of history · worst day 0%

91 days agonow

100.0%

Uptime 24h

91 of 91 checks

Tools

read from the server

137 ms

Response time

average over 24h

authentication required

Access

streamable-http

Nothing serious here today

Today is the operative word: we check Library every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.

Connect this server

Endpoint below is the one we actually reach during checks — not the one copied from a README. Last verified 11 min ago.

Claude Code Claude Desktop Codex Cursor VS Code

run in your terminal

claude mcp add library --transport http https://library.siftdex.com/mcp

Settings → Connectors → Add custom connector, then paste this address

https://library.siftdex.com/mcp

~/.codex/config.toml

[mcp_servers.library]
url = "https://library.siftdex.com/mcp"

.cursor/mcp.json

{
  "mcpServers": {
    "library": {
      "url": "https://library.siftdex.com/mcp"
    }
  }
}

.vscode/mcp.json

{
  "servers": {
    "library": {
      "type": "http",
      "url": "https://library.siftdex.com/mcp"
    }
  }
}

This endpoint answered with an authorization challenge. The server is running, but it did not say what kind of credentials it expects.

Available tools 13

Read directly from the server with tools/list, grouped by what they act on. If a tool disappears, we record the date.

changed

changed_since

Net added/modified/removed pages since a cursor (run_id from snapshot_status, or ISO ts), from the hash-chained changelog. Returns a fresh cursor; up_to_date:true = nothing new. Read only what it returns instead of re-reading the corpus.

diff

diff_md

Unified diff of ONE page between two published snapshots. from=<run_id|ISO ts> required; to defaults to current. Changed hunks + both content_hashes only — pair with changed_since for which pages moved.

dir

list_dir

Immediate children of current/<path>: name, kind, size. Up to 500.

facts

read_facts

Read one facts/<schema>/*.json (numbers/rates/thresholds). 20K-char cap.

fetch

fetch

Fetch one document by search id (<index>:<path>). Returns full text + metadata receipt (content_hash, fetched_at, run_id) — prove(url) upgrades it to an offline-verifiable proof.

glob

glob_corpus

fnmatch glob (e.g. md/**/*.md). Cursor-paginated. 500 paths/page.

grep

grep_corpus

Regex over current/md. files_only=true skips snippets. Capped 200 hits. INVALID_PATH/TIMEOUT.

indexes

list_indexes

The library catalog: slug, displayName, source, pageCount, lastPublishedAt. Other tools take optional index=<slug>.

manifest

query_manifest

Read-only SELECT against manifest's public columns (see corpus.contract §3.6). format='columnar' saves ~40% bytes. Cursor-paginated. 500 rows/page.

prove

prove

Merkle inclusion proof that url's content_hash is committed by the snapshot root (+ RFC-3161 token). as_of=<run_id|ISO ts> proves a past snapshot. included:false = not an error. Verify offline: python -m sift.verify_proof.

read

read_md

Read one markdown file (frontmatter + body slice). verify=true recomputes content_hash: pass → verified:true, mismatch → VERIFY_MISMATCH.

search

Library-wide search (ChatGPT connector schema). query → results[{id,title,url,text}]; pass a result id to fetch. Native tools (grep_corpus, read_md) offer finer control.

snapshot

snapshot_status

Published snapshot: run_id, gates, integrity. Call first. Pre-publish → NO_PUBLISHED_RUN.

Endpoints

URL	Transport	State	Latency	Checked
https://library.siftdex.com/mcp	streamable-http	needs auth	108 ms	11 min ago

Alternatives to Library

same job, measured the same way

Context

by neuledge

Local-first documentation for AI agents. Offline, instant, private doc retrieval.

1 781 installs/wk local only

Kronicle Lite

by rafe-walker

Hash-chained local document store with signed receipts and an orient protocol for stale agent memory

2 793 installs/wk local only

Pactlyst.ai — PactCheck document review for agents

by fikpeze123

Score documents and check agent work — signed PactCheck Receipts: proof-of-review for agents.

answering

Oh Fax

by ohfax

Outbound US/CA fax for agents: quote PDF, pay via MPP, get delivery receipt.

4 tools answering

SuperLocalMemory

by qualixar

Local-first memory for AI agents: recall with sources, answer checks, documents, mesh, governance.

1 251 installs/wk local only

Dibri

by dibri

Verified answers for AI agents: only from approved documents, with sources, or a refusal.

1 tools answering

Elliot

by emanuelecell4

Independent verifier for AI agents: checks agent claims and doc promises against the code.

local only

Skills

by tasklinkr

Executable skill packages for agents: vetted competence, ready to run, not just documents.

1 tools answering

Library — questions

Answers built from our own checks of this server.

What can Library do?

It exposes 13 tools, read directly from the server on our last check. Among them: changed_since, diff_md, fetch, glob_corpus, grep_corpus, list_dir and 7 more. The full list with descriptions is on this page — we take it from the server itself via tools/list, not from a README. How MCP servers expose tools in the first place →

Is Library working right now?

We send a real MCP handshake every 15 minutes. Over the last 24 hours 91 of 91 checks got a reply (100.0%), average response time 137 ms. The bar chart above shows every period we have measured.

How do I connect Library?

Copy the ready config from this page. We generate it for Claude Code, Claude Desktop, Codex, Cursor and VS Code, each in the form that client actually reads. It is a remote server, so there is nothing to install — the client connects to the address. You will need your own credentials: it refuses anonymous clients.

Does Library need an API key?

Yes. Every time we knock, Library answers with an authorization challenge instead of its tool list — that is how we know it is running and gated rather than broken. Bring your own credentials and it will talk.

How fast is Library?

It answers our handshake in 137 ms on average, which is faster than 86% of all working MCP servers we measure. That puts it in the quick quarter of the ecosystem. The comparison comes from our own checks across the whole registry, every 15 minutes.